This month, I've been thinking about hospitality - not the industry, but the practice. What it means to anticipate needs, go further than reasonable, and genuinely serve the people who trust you. Turns out, that same principle applies whether you're running a restaurant or building technology.
We're deep in Q1 planning right now, setting 2026 roadmaps. And I keep coming back to that hospitality question: how do we build things that anticipate what people need before they even know to ask? How do we show appreciation to our most dedicated users?
With Thanksgiving this week, it feels like the right moment to reflect on both what we're building and who we're building it with. Here's what's on my mind this month.
This Month's Take
Fifteen years ago, I was sitting at Le Bernardin, living my dream of visiting the world's best restaurants. Across from me, a group of six settled in, dressed to the nines, scanning their menus in the restaurant's dim lighting.
The maître d' approached quietly with a small cigar box. He opened it just under table height next to one of the men, who visibly grinned as he pulled out a pair of reading glasses. The entire table delighted, quickly asking for their own pairs.
That moment set my love for hospitality - and a rule I think about when building search today. Hospitality isn't about delivering great service. It's about anticipating needs and going just a little bit further than feels reasonable. Those reading glasses in dim lighting? That's hospitality at its best.
Now think about the last time you used a website's search box. You probably got a long, long list of results ranked by some algorithm, maybe a "no results found" message if your phrasing was slightly off. You were on your own to figure out what you actually needed.
This is the gap I keep thinking about: your highest-intent visitors are getting the least hospitality.
And it matters more than ever right now. When Google's AI Mode rolled out, it changed how people find information overnight. Now, when someone searches, they get an AI summary answer, with no clickthrough back to you.
The opportunity is enormous. In user interviews, people consistently tell us they turn to Google because on-site search is so frustrating to use. Your visitors want to stay with you - if you make it easy for them.
Read this if you’ve struggled with making site search a first-class part of your site.
From Our Work
Everyone wants to move fast. Ship features, test ideas, iterate quickly. But when you're building AI that people trust with important questions, speed without safety is reckless.
The answer? Robust evaluations.
This is the unglamorous work that makes everything else possible. Before we launch anything, we build custom benchmarks testing exactly what matters for that partner. Not generic AI benchmarks, but personalized queries designed to find and test the boundaries—like knowing when a parent's question crosses from general sleep advice to specific medical guidance that needs a doctor.
We test hundreds of questions. We verify every citation. We stress-test edge cases. And we keep testing after launch, monitoring responses to catch issues before users do.
This is how you treat users with respect. Not by moving fast and breaking things, but by building systems worthy of their trust.
Speed matters. But accuracy matters more.
Read this if you're building AI systems where trust isn't optional.
What We're Building
Public policy decisions affect millions of lives. Getting them right requires synthesizing decades of research, comparing intervention effectiveness, and understanding nuanced tradeoffs.
The County Health Rankings & Roadmaps team has been doing this work for years through What Works for Health - a comprehensive database of evidence-based strategies for improving community health. Over 400 strategies, each carefully evaluated for effectiveness and equity impact.
The challenge? All that expertise can be hard to navigate. Public health professionals knew it was there, but finding the right strategy for their specific situation meant clicking through dozens of pages.
We built something better together. Now when someone asks "What works for reducing childhood obesity in rural communities?" they get targeted, evidence-based strategies with clear effectiveness ratings - all cited back to What Works for Health's original research.
This is hospitality in action. Taking deep expertise and making it genuinely accessible to the people who need it most.
Read this if you're working on projects where deep expertise should reach everyone who needs it.
Gratitude: Idea Worth Stealing
My friend Andres Glusman at DoWhatWorks shared a Thanksgiving tradition with me that I immediately wanted to steal. This week, we're trying it with the Dewey team for the first time.
Here's how it works: Someone sends a form to the team where everyone anonymously shares what they appreciate about each person. One person compiles the responses into lists. Then, at your team meeting, you randomly assign people to read each other's lists out loud.
Andres says it simply: "It's overwhelming to hear what each person appreciates about you and moving to see your teammates recognized. Bring tissues."
DoWhatWorks has done this three years running. Everyone looks forward to it. If you're looking for a way to close out the year with your team, steal this.
One Question for You
Whose work do you appreciate, that deserves more attention for what they're building or creating?
If you're also in planning mode and exploring how AI fits into your 2026 strategy, this is a great time to connect. We're talking to organizations about partnerships now, before holiday calendars fill up.
Grab time on my calendar or respond here and we’ll chat.
See you in December,
Alex
You're receiving this because you signed up with Dewey Labs - where we build expert-first AI tools. We work with publishers, brands, researchers, and non-profits to create search and Q&A experiences grounded in trust and accuracy. We’ll be sending a monthly dispatch on the future of the expert economy like this one, along with occasional product updates.